Miniaturization technologies for cost-effective AmpliSeq library preparation for next generation sequencing

Purpose AmpliSeq technology, the target enrichment method for next-generation sequencing (NGS), enables quick and easy detection of the genomic “hot spot” region frequently mutated in species. Even though the cost of NGS has decreased, library preparation cost accounts for a more significant proportion of the total cost. If AmpliSeq library can be prepared at a lower cost, large-scale precision oncology can be more easily carried out. Furthermore, this technology can be widely applied not only to medical research, but also to polymorphism detection in biology. This study aimed to reduce the cost of AmpliSeq library preparation by adopting miniaturization technology. Methods We used approximately 10 ng of genomic DNA for ultra-multiplex PCR of 384, 768, 1152, 1920, and 3072 amplicons. Multiplex PCR was performed in a total volume of 1.6, 2.0, and 2.4 μL, using a nano-liter liquid handler, for library preparation. Results The success rate of library construction decreased with decreasing total multiplex PCR reaction volume. Using 1.6-, 2.0-, and 2.4-μL reactions, the success rates of ultra-multiplex PCR were 25%, 95%, and 100%, respectively. We could stably create libraries of the correct amplicon size, with an amplicon number of approximately 1500 or less. As a result of NGS, uniformity of PCR amplification and read length of quality-checked libraries were hardly affected by the number of amplicons. Conclusion Here, we show that the minimum volume for a stable reaction was 2.4 μL and the maximum number of amplicons obtained was approximately 1500. The protocol saved 86.8% in reagent usage and reduced handling time by 85% compared to that required by the manual protocol. Therefore, miniaturization technologies could reduce the cost of AmpliSeq library preparation through minimization of reagents.


Introduction
Genotyping by sequencing (GBS), using next-generation sequencing (NGS), is becoming an increasingly useful technique in the fields of biology and agronomy [1][2][3]. In recent years, the cost of sequencing per base has drastically reduced, although the library preparation cost and time involved still remain high.
AmpliSeq technology (Thermo Fisher Scientific, Waltham, MA) is a GBS method that is often used in cancer research [4]. Although the AmpliSeq library kit is very expensive (about 64 USD or 9,500 JPY/sample), highly multiplexed PCR (of approximately 6000 primer pairs) is made possible with it.
AmpliSeq technology is beginning to be used in other fields as well and is sold as AgriSeq in the fields of biology and agronomy [5,6].
In the field of agronomy, especially in crop-breeding sites, genotyping of thousands of large samples is carried out often. However, GBS is still expensive, requiring a considerable amount of time and labor for analysis. It is difficult to apply it for practical crop breeding, and thus, is mainly used in research or project. A quick way to decrease the cost of GSB is to decrease library preparation cost. The simplest way for that is to reduce the reaction volume. In manual library preparation, volume reduction by more than half to one quarter is not realistic. Quality deterioration of the library is highly probable with reduction in the amount of reaction volume in case of manual preparation. Therefore, to sufficiently reduce the cost, it is necessary to have a robot capable of transferring minute quantities of liquid that cannot otherwise be handled by humans.
Till date, only few protocols for miniaturized library preparation in GBS have been reported. To reduce the cost and time limitations of current library preparation techniques, we tried to adapt our AmpliSeq, one such GBS method that follows a miniaturized reagent protocol, using Mosquito HTS (TTP Labtech, Royston, UK), which is a positive displacement pipetting instrument. Mosquito HTS offers highly accurate and precise multichannel pipetting (8 or 16 channels) from 25 nL to 1.2 µL. Mosquito HTS was originally used for protein crystallization [7]. Recently, Mosquito HTS and another model Mosquito HV have been used for single-cell RNASeq analysis [8]. Mosquito HV being a high-volume model (500 nL to 5 µL) does not save enough reagent and does not minimize the cost of AmpliSeq library preparation.
Here, we describe a method that allows miniaturized, automated, and cost-and time-efficient 384-well library preparation with its quality and performance.
The in-silico amplicon size distributions are shown in Figure 1. The peak size in all panels was approximately 280 bp ( Figure 1).

AmpliSeq library preparation with Mosquito HTS
NGS library was constructed using the Ion AmpliSeq Library Kit 2.0 (Thermo Fisher Scientific, Waltham, MA) [4]. To miniaturize this procedure for Mosquito HTS, we tested a modified protocol in reduced volume of reagent (Table 1). Although making a master mix reduces the number of dispensing steps, a dead volume is likely to occur. Therefore, template DNA and each reagent were manually dispensed into each row of 384-well Low Volume Serial Dilution (LVSD) plates [source plate] for automated liquid   The mixed adapter-ligated libraries were purified using 1.5-fold volume of AMPure XP Reagent, followed by the addition of 150 μL of freshly prepared 70% ethanol to each library. The washing step was repeated twice, the ethanol was completely removed, and beads were air-dried for 3−5 min, while the plate was on the magnetic rack. The library was eluted from the beads with 23 μL low TE; 20 μL of the supernatant was transferred to a clean tube.

7
The concentration and size of amplicons were determined using D1000 ScreenTape with Agilent 4200 TapeStation (Agilent Technologies, Santa Clara, CA) according to the manufacturer's instruction. After quantification, mixed library was diluted to a concentration of 100 pM prior to template preparation.
Next indicating that the multiplex PCR did not work well in miniaturized PCR reaction volumes of 2 µL or less.
The quality-checked libraries were pooled into the appropriate concentration, and the mixed library was diluted to a concentration of 100 pM prior to template preparation as described above.

Sequencing and coverage analysis
Final libraries were sequenced on the IonTorrent S5 system (Thermo Fisher Scientific, Waltham, MA).
Template preparation consisting of emulsion PCR, enrichment of beads containing the template, and chip loading, was performed with the Ion Chef instrument and Ion S5 Kit-Chef according to the manufacturer's instruction (Thermo Fisher Scientific, Waltham, MA). After the preparation of ion sphere particles (ISPs), sequencing for 500 cycles was performed with an Ion Torrent Ion S5 system using Ion   We obtained an average depth coverage of 672 (a total of 65.9 M reads) in four samples with 384, 768, 1152, 1920, and 3072 amplicon panels. A similar trend was observed for a low % of on-target rate and uniformity in the 1152 amplicon panel. These results indicated that the differences in on-target rate and uniform PCR amplification are panel-dependent and that miniaturized volumes of reagents do not affect on-target rate or uniformity. Compared to that obtained with the standard 20-μL volume, mean read length, on-target rate, and uniformity of the miniaturized protocol were not different (Figure 3).  (Table 3).
Similarly, the automated system for library preparation was time-saving. Making libraries of 384 samples by hand, at the same time, with the manual protocol, would take approximately 66 min.
Although Mosquito HTS would take almost similar time (approximately 44 min) to make 384 libraries, we do not need much labor in the automated system. Since it takes time to mix reagents by pipetting, followed by individual reagent transfer (no master mix), it is difficult to shorten the library preparation time; however, the handling time can be greatly reduced due to the automated pipetting. Mosquito HTS has two models: 2-way and 5-way (2 or 5 plates can be set at once). Using the Mosquito HTS 5way model, labor can be further minimized as some plates can be set at once and processed continuously. Alternatively, although dead volume may be generated when a master mix is prepared, dispensing time is shortened. Dead volume has little effect on the cast of library prep when dealing with sample sizes of 1000 or more.

Discussion
In this study, we presented a miniaturization protocol for AmpliSeq library preparation. Compared to the standard protocol, our protocol is inexpensive and not labor-intensive. Empirically, stable results were obtained when the reaction volume was 2.4 µL and the number of amplicons was 1500 or less. The influence of the difference in uniformity and amplicon size of multiplex PCR was greater than the influence of the type of thermo-cycler used and reaction volume. Miniaturization seemed to make the protocol easier to be influenced by environmental factors.
This is the first application of the Mosquito HTS for the AmpliSeq ultra-multiplex PCR protocol and adds to other existing protocols for NGS library preparation [8][9][10][11][12]. However, there are two limitations of this Mosquito protocol. First, Mosquito HTS should be set in a cold room, whereas in this study, we used it at room temperature. Moreover, we always used freshly prepared reagents. The reagent was dispensed in advance into 384-well plate (source plate) before use, and the residual reagent was stored at -20 ºC to be used in the next experiment. As the number of times of use and time lapse increased, the quality of the library declined in the subsequent experiments (lower uniformity, data not shown). This could be prevented by using the Mosquito HTS in a cold room.
Second, Mosquito HTS cannot not be used for bead clean-up. Therefore, it is necessary to use another conventional dispensing robot. As an alternative method, a two-step process was used to optimize the pooling of the samples. First, all the libraries were mixed equally and sequence-skimmed; then, the concentration from the first sequencing result was adjusted and sequenced again. Finally, the initial cost of the robot is high. However, these robots are very useful and greatly reduce the experimental cost if housed in core labs or core facilities.
In conclusion, we presented a protocol to prepare sequencing libraries in miniaturized volume using AmpliSeq library kit (Thermo Fisher Scientific, Waltham, MA). With this protocol, it is possible to prepare 384 libraries with just 12% of the standard reagent volume, at less than 13.1% of the cost, and in less than 18.2% of the time required in the standard manual protocol. This should help the advancement of not only clinical genomics, but also large-scale genotyping in the agronomic field and projects such as AgriSeq.